Horizontal machining center hydraulic braking system

Abstract

This application relates to a hydraulic braking system for a horizontal machining center, specifically in the technical field of hydraulic braking for CNC machine tools. It includes a braking prediction module, a tiered hydraulic valve group, and a digital twin control module. The braking prediction module analyzes the time series of spindle speed, load current, and temperature data to predict the braking initiation point. The tiered hydraulic valve group consists of a proportional servo valve and a piezoelectric ceramic auxiliary valve. The digital twin control module, based on a computational fluid dynamics simulation model and a PID controller, simulates the hydraulic circuit and generates an optimal pressure curve to control the tiered hydraulic valve group. This application uses the braking prediction module to monitor and analyze spindle operating parameters in real time, enabling it to predict braking needs and determine the braking initiation point in advance, thereby avoiding braking timing deviations caused by response lag in traditional braking systems.

Description

Technical Field

[0001] This invention relates to the field of hydraulic braking technology for CNC machine tools, and in particular to a hydraulic braking system for a horizontal machining center. Background Technology

[0002] Currently, horizontal machining centers face increasingly stringent requirements for braking system response speed, stability, and energy efficiency during high-speed, heavy-load machining. Traditional hydraulic braking systems often employ single proportional valve control, achieving braking through a fixed pressure curve. While this can meet general machining needs, it is prone to problems such as pressure overshoot and vibration under highly dynamic operating conditions. In recent years, with the development of CNC machine tools towards higher speeds and precision, the industry has begun to explore the introduction of predictive control, digital twins, and other technologies to optimize the braking process, but a mature integrated solution has yet to be formed.

[0003] In existing technologies, conventional methods include: ① using purely mechanical brakes, which achieve braking through friction pad contact. The advantage is simple structure, but there are problems such as severe wear and poor heat dissipation; ② using hydraulic systems controlled by ordinary proportional valves, which can change pressure by adjusting the current, but the response time is usually more than 50ms and cannot suppress pressure fluctuations; ③ some high-end equipment uses servo valves to replace proportional valves, which can improve the response speed, but the cost is high and the requirements for oil cleanliness are extremely high.

[0004] Regarding the aforementioned technologies, the purpose of this application is to solve the problems of slow response, braking vibration, and high energy loss in traditional hydraulic braking systems under high-speed and heavy-load processing scenarios. Summary of the Invention

[0005] The purpose of this application is to provide a hydraulic braking system for a horizontal machining center to solve the problems of slow response, braking vibration and high energy loss of traditional hydraulic braking systems in high-speed and heavy-load machining scenarios.

[0006] The hydraulic braking system for a horizontal machining center provided in this application adopts the following technical solution: A hydraulic braking system for a horizontal machining center includes a braking prediction module, a tiered hydraulic valve group, and a digital twin control module. The braking prediction module analyzes the time series of spindle speed, load current, and temperature data to predict the braking initiation point. The tiered hydraulic valve group consists of a proportional servo valve and a piezoelectric ceramic auxiliary valve. The digital twin control module, based on a computational fluid dynamics simulation model and a PID controller, simulates the hydraulic circuit and generates an optimal pressure curve to control the tiered hydraulic valve group. The braking prediction module integrates a spindle speed sensor, a load current detector, and a temperature probe, all of which are bolted to corresponding interfaces on the spindle housing. The proportional servo valve is connected to the hydraulic pump outlet pipeline via a flange. The piezoelectric ceramic auxiliary valve is threaded to a downstream branch of the main valve. The industrial control computer of the digital twin control module is mounted in an electrical cabinet via a bracket, and its signal lines are connected to the sensors and valves using shielded twisted-pair cables.

[0007] By adopting the above technical solutions, the braking prediction module can capture braking signals in advance and predict the braking start point, reserving response time for braking actions and solving the problem of lag in traditional systems. The staged hydraulic valve group, combined with the control of the proportional servo valve and the rapid response of the piezoelectric ceramic auxiliary valve, can achieve pressure regulation. At the same time, the digital twin control module generates the optimal pressure curve through simulation and, together with the PID controller, controls the valve group in real time, which can suppress pressure overshoot and fluctuation, reduce braking jitter, and avoid unnecessary energy loss.

[0008] Preferably, the braking prediction module further includes a CNC code parser, used to parse the machining code segment to be executed in real time, extract geometric features and cutting parameters. The braking prediction module adopts a multimodal feature fusion network to fuse the temporal feature vector output by the LSTM neural network with the code feature vector output by the CNC code parser, and performs multi-task learning. By adopting the above technical solution, the braking prediction module uses a multimodal feature fusion network to fuse the time-series feature vectors such as spindle speed, load current, and temperature output by the LSTM neural network with the code feature vectors such as machining code geometric features and cutting parameters parsed in real time by the CNC code parser. This enables the prediction of the braking start point, allowing the braking system to prepare for response in advance, thus solving the problem of response lag caused by the lack of prediction in traditional braking systems.

[0009] Preferably, the digital twin control module uses intrinsic orthogonal decomposition or dynamic mode decomposition to reduce the dimensionality of the simulation model to obtain a surrogate model. The digital twin control module generates braking pressure curve plans for various working conditions offline based on the surrogate model and clusters the plan curves. By adopting the above technical solutions, the intrinsic orthogonal decomposition or dynamic mode decomposition model reduction technology can reduce the complexity of the simulation model, reduce the model calculation time, and improve the response efficiency of the digital twin control module. At the same time, based on the surrogate model, offline generation and clustering of multi-condition pressure curve plans can reserve the optimal control scheme under different processing scenarios in advance, avoid the delay of online real-time simulation, further optimize the braking response speed, and reduce the computational energy consumption in the control process.

[0010] Preferably, during online control, the optimal pre-plan in the pre-plan library is quickly matched based on the current operating parameters using the K-nearest neighbor algorithm to achieve microsecond-level initialization of the pressure curve; By adopting the above technical solution, the K-nearest neighbor algorithm can match the optimal braking pressure curve from the clustered pre-plan library based on the current spindle speed, load, temperature and other operating parameters, achieving microsecond-level initialization. Compared with the response time of more than 50ms in traditional systems, this shortens the braking preparation time. At the same time, the rapid initialization of the optimal pre-plan allows the braking system to be in the optimal control state from the initial stage of startup, reducing fluctuations and overshoot during pressure regulation and lowering the probability of braking jitter.

[0011] Preferably, the piezoelectric ceramic auxiliary valve is further provided with a deformation micro-sensor. The feedback signal of the deformation micro-sensor is connected to the digital twin control module to sense the instantaneous viscosity change of the oil and realize cross-sensing compensation. A model predictive controller is also provided between the digital twin control module and the staged hydraulic valve group. The model predictive controller uses a surrogate model as the internal predictive model, solves the finite time domain optimization problem in each control cycle, obtains the optimal control sequence, and sends it to the PID controller for execution. By adopting the above technical solutions, the deformation micro-sensor can capture the instantaneous viscosity change of the oil in real time and feed it back to the digital twin control module, avoiding pressure regulation deviation caused by oil viscosity changes, improving the stability of braking control, and thus reducing braking vibration caused by viscosity fluctuations; the model predictive controller, based on the surrogate model, predicts the pressure change trend of each control cycle in advance, solves the optimal control sequence and sends it to the PID controller, further suppressing pressure fluctuations and improving braking accuracy.

[0012] Preferably, the staged hydraulic valve group further includes a miniature variable displacement hydraulic motor / pump unit integrated in the hydraulic circuit. The miniature variable displacement hydraulic motor / pump unit is used to convert hydraulic energy into electrical energy for storage during braking and to assist in outputting power when the system needs to accelerate or compensate for pressure. The digital twin control module is also used to optimize the energy recovery and utilization of the miniature variable displacement hydraulic motor / pump unit according to the predicted processing cycle. By adopting the above technical solution, the miniature variable displacement hydraulic motor / pump unit can recover excess hydraulic energy during braking, thereby solving the problem of high energy loss in traditional braking systems and reducing overall energy consumption.

[0013] Preferably, the digital twin control module further includes a thermo-mechanical coupling model, which is used to couple the thermal deformation model of the spindle and bearing with the mechanical model of the hydraulic braking system. The hydraulic braking system also includes an online parameter identification module, which recursively estimates and updates the key parameters in the digital twin model based on the actual collected vibration, temperature and pressure data. By adopting the above technical solutions, the thermo-mechanical coupling model can combine the thermal deformation of the spindle and bearings with the hydraulic braking system, making the digital twin simulation closely match the actual working conditions. This avoids problems such as changes in brake clearance and pressure adjustment deviations caused by thermal deformation, further reducing brake vibration. At the same time, the online parameter identification module dynamically updates the key parameters of the digital twin model based on real-time collected vibration, temperature, and pressure data, avoiding simulation and actual deviations caused by parameter drift during long-term operation.

[0014] Preferably, the online parameter identification module uses an extended Kalman filter, which uses key model parameters such as hydraulic oil viscosity-temperature coefficient and bearing equivalent thermal resistance as part of the state variables to construct an extended state space model. Through state prediction and update steps, the model parameters are recursively estimated and updated using actual observation data. By adopting the above technical solution, the extended Kalman filter can capture the dynamic changes of key parameters such as hydraulic oil viscosity-temperature coefficient and bearing equivalent thermal resistance through its excellent nonlinear estimation capability. These parameters are then incorporated into the extended state-space model for prediction and updating, thereby tracking parameter changes in real time and correcting model deviations. By updating key parameters, the consistency between the digital twin model and the actual braking system is ensured, thereby suppressing pressure fluctuations and braking vibrations caused by parameter deviations.

[0015] Preferably, it also includes a vibration suppression module and an adaptive damping adjustment unit. The vibration suppression module is equipped with a triaxial acceleration sensor and an adaptive notch filter. The vibration suppression module is used to monitor the vibration frequency and amplitude during the spindle braking process, and actively suppresses the resonance point by adjusting the compensation pulse width of the piezoelectric ceramic auxiliary valve.

[0016] Preferably, the vibration suppression module collects vibration signals during the braking process through a triaxial acceleration sensor installed at the bottom of the spindle box, converts the time-domain signal into frequency-domain features using Fourier transform, identifies the resonant frequency in the 10-500Hz frequency band, and then controls the magnetorheological damper in the adaptive damping adjustment unit to change the magnetic field strength and adjust the damping coefficient in real time. Beneficial effects

[0017] In summary, this application includes at least one of the following beneficial technical effects: This invention provides a hydraulic braking system for a horizontal machining center. By incorporating a CNC code parser and a multimodal feature fusion network into the braking prediction module, it integrates time-series data with machining code features, improving the prediction accuracy of the braking start point and reserving response time for the braking system, thereby increasing the response speed. Simultaneously, the pressure regulation of the staged hydraulic valve group and the optimal pressure curve and real-time control generated by the digital twin control module can suppress pressure fluctuations and braking jitter, avoiding the machining accuracy degradation caused by response lag and pressure overshoot in traditional systems.

[0018] This invention provides a hydraulic braking system for a horizontal machining center. By setting a deformation micro-sensor on the piezoelectric ceramic auxiliary valve to achieve cross-sensing compensation, and coordinating the control of the model predictive controller and the PID controller, the stability and accuracy of braking control are improved. With the addition of a micro variable displacement hydraulic motor / pump unit, a thermo-mechanical coupling model and an extended Kalman filter, the system can not only further improve its adaptive capability under complex working conditions, but also reduce braking deviations caused by changes in oil viscosity, thermal deformation and parameter drift, thereby reducing energy loss and improving the overall energy efficiency ratio of the system.

[0019] This invention provides a hydraulic braking system for a horizontal machining center. Through a proxy model constructed by a digital twin control module and an offline pre-plan library, the braking process can be digitally rehearsed and responded to quickly. Combined with real-time updates of key parameters such as the hydraulic oil viscosity-temperature coefficient and bearing equivalent thermal resistance by an online parameter identification module, the digital twin model maintains dynamic consistency with the actual system. Thus, during the braking initiation phase, the optimal pressure curve pre-plan is matched in microseconds using the K-nearest neighbor algorithm. Furthermore, during braking, the control sequence is continuously optimized by a model predictive controller, enabling the proportional servo valve and piezoelectric ceramic auxiliary valve to work collaboratively, suppressing pressure overshoot and fluctuations. Attached Figure Description

[0020] Figure 1 This is a system flowchart of the present invention. Detailed Implementation

[0021] The following is in conjunction with the appendix Figure 1 This application will be described in further detail below.

[0022] Example 1: A hydraulic braking system for a horizontal machining center, referring to... Figure 1The system includes a braking prediction module, a tiered hydraulic valve group, and a digital twin control module. The braking prediction module employs an LSTM neural network with gating mechanisms including a forget gate, input gate, and output gate. It analyzes the time series of spindle speed, load current, and temperature data to predict the braking initiation point. The tiered hydraulic valve group consists of a proportional servo valve and a piezoelectric ceramic auxiliary valve. The proportional servo valve provides the reference pressure, while the piezoelectric ceramic auxiliary valve performs fine-tuning compensation. The digital twin control module, based on a computational fluid dynamics simulation model and a PID controller, simulates the hydraulic circuit and generates an optimal pressure curve every 5ms to control the tiered hydraulic valve group. The braking prediction module integrates a spindle speed sensor, a load current detector, and a temperature probe. The heads are bolted to the corresponding interfaces of the spindle housing. The proportional servo valve is connected to the hydraulic pump outlet pipeline via a flange. The piezoelectric ceramic auxiliary valve is threaded to the downstream branch of the main valve. The industrial control computer of the digital twin control module is mounted in the electrical cabinet via a bracket. Its signal lines are connected to each sensor and valve using shielded twisted-pair cables. The braking prediction module can capture braking signals in advance and predict the braking start point, reserving response time for braking action and solving the problem of response lag in traditional systems. The staged hydraulic valve group, combined with the control of the proportional servo valve and the rapid response of the piezoelectric ceramic auxiliary valve, can achieve pressure regulation. At the same time, the digital twin control module generates the optimal pressure curve through simulation and, together with the PID controller, controls the valve group in real time, which can suppress pressure overshoot and fluctuation, reduce braking jitter, and avoid unnecessary energy loss.

[0023] The braking prediction module also includes a CNC code parser, which is used to parse the machining code segment to be executed in real time, extract geometric features and cutting parameters. The braking prediction module adopts a multimodal feature fusion network, which fuses the temporal feature vector output by the LSTM neural network with the code feature vector output by the CNC code parser through an attention mechanism, and performs multi-task learning. Specifically, multi-task learning includes predicting the braking initiation point, the probability distribution of braking type, the initial pressure threshold, and its ideal rate of change. The braking prediction module uses a multimodal feature fusion network to fuse the temporal feature vectors such as spindle speed, load current, and temperature output by the LSTM neural network with the code feature vectors such as the geometric features and cutting parameters of the machining code parsed in real time by the CNC code parser. This enables the braking initiation point to be predicted, allowing the braking system to prepare for response in advance, thereby solving the problem of response lag caused by the lack of prediction in traditional braking systems.

[0024] The digital twin control module uses intrinsic orthogonal decomposition or dynamic mode decomposition to reduce the dimensionality of the simulation model, obtaining a surrogate model. The input of the surrogate model is the operating condition parameter vector. The output is a pressure curve for a future period. The digital twin control module generates braking pressure curve plans for various operating conditions offline based on a proxy model. The system employs K-means or hierarchical clustering algorithms to cluster the pre-plan curves, selecting the central curve of each cluster as a representative pre-plan and storing it in the pre-plan library. A fast lookup table for parameter X to pre-plan index is also established. Intrinsic orthogonal decomposition or dynamic mode decomposition model reduction techniques can reduce the complexity of the simulation model, reduce model computation time, and improve the response efficiency of the digital twin control module. At the same time, based on the surrogate model, offline generation and clustering of multi-condition pressure curve pre-plans can reserve the optimal control scheme under different processing scenarios in advance, avoid the delay of online real-time simulation, further optimize braking response speed, and reduce computational energy consumption during the control process.

[0025] During online control, the optimal pre-plan in the pre-plan library is quickly matched based on the current operating parameters using the K-nearest neighbor algorithm, achieving microsecond-level initialization of the pressure curve. Specifically, the digital twin control module uses the K-nearest neighbor algorithm to find the k most similar pre-plans in the pre-plan library based on the current operating parameters, and weights them according to similarity to generate the initial optimal pressure curve, achieving microsecond-level initialization. The K-nearest neighbor algorithm can match the optimal braking pressure curve from the clustered pre-plan library based on the current spindle speed, load, temperature, and other operating parameters, achieving microsecond-level initialization. Compared to the response time of more than 50ms in traditional systems, this shortens the braking preparation time. At the same time, the rapid initialization of the optimal pre-plan allows the braking system to be in the optimal control state from the initial start-up, reducing fluctuations and overshoot during pressure regulation and lowering the probability of braking jitter.

[0026] The similarity metric uses a weighted Euclidean distance: Among them, weight The sensitivity of each parameter to the braking process is determined by its influence.

[0027] The piezoelectric ceramic auxiliary valve is also equipped with a deformation micro-sensor. The feedback signal from the deformation micro-sensor is connected to the digital twin control module to sense the instantaneous viscosity change of the oil and achieve cross-sensing compensation. A model predictive controller is also installed between the digital twin control module and the staged hydraulic valve group. The model predictive controller uses a surrogate model as its internal predictive model to solve the finite-time domain optimization problem in each control cycle, obtain the optimal control sequence, and send it to the PID controller for execution. Specifically, the formula for calculating the optimal control sequence is: ; In the formula, To predict the time domain, To control the time domain, For the reference pressure trajectory from the digital twin, This is the weight matrix. To control the change in the variable, after obtaining the optimal control sequence, the first control variable is sent to the PID controller for execution. The deformation micro-sensor can capture the instantaneous viscosity change of the oil in real time and feed it back to the digital twin control module, avoiding pressure regulation deviation caused by oil viscosity changes, improving the stability of braking control, and thus reducing braking vibration caused by viscosity fluctuations. The model predictive controller is based on the surrogate model, predicts the pressure change trend of each control cycle in advance, solves the optimal control sequence and sends it to the PID controller, further suppressing pressure fluctuations and improving braking accuracy.

[0028] The staged hydraulic valve manifold also includes a miniature variable displacement hydraulic motor / pump unit integrated into the hydraulic circuit. This unit converts hydraulic energy into electrical energy for storage during braking and provides auxiliary power output when the system needs acceleration or pressure compensation. The recovery power of the miniature variable displacement hydraulic motor / pump unit is... ,in, For efficiency, Where n is the pressure difference, n is the rotational speed, and D is the displacement, the digital twin control module is also used to optimize the energy recovery and utilization of the micro variable displacement hydraulic motor / pump unit according to the predicted processing cycle. The micro variable displacement hydraulic motor / pump unit can recover excess hydraulic energy during braking, thereby solving the problem of high energy loss in traditional braking systems and reducing overall energy consumption.

[0029] The digital twin control module also includes a thermo-mechanical coupling model, which couples the thermal deformation model of the spindle and bearings with the mechanical model of the hydraulic braking system. The hydraulic braking system also includes an online parameter identification module, which recursively estimates and updates the key parameters in the digital twin model based on the actual collected vibration, temperature, and pressure data after each braking operation. The thermo-mechanical coupling model can combine the thermal deformation of the spindle and bearings with the hydraulic braking system, making the digital twin simulation closely match the actual working conditions and avoiding problems such as brake clearance changes and pressure adjustment deviations caused by thermal deformation, further reducing braking vibration. At the same time, the online parameter identification module dynamically updates the key parameters of the digital twin model based on the real-time collected vibration, temperature, and pressure data, avoiding simulation and actual deviations caused by parameter drift during long-term operation.

[0030] The online parameter identification module employs an extended Kalman filter (EPF). The EPF incorporates key model parameters such as the hydraulic oil viscosity-temperature coefficient and the equivalent thermal resistance of the bearing as part of the state variables, constructing an extended state-space model. Through state prediction and update steps, it recursively estimates and updates the model parameters using actual observation data. Specifically, the recursive update steps include: State prediction: ; Covariance prediction: ; Kalman gain: ; Status Update: ; Covariance update: ; In the formula, state x includes pressure, temperature, vibration, and parameters to be identified, while f and h are the state transition and observation functions containing thermo-mechanical coupling relationships, respectively. The extended Kalman filter can capture the dynamic changes of key parameters such as hydraulic oil viscosity-temperature coefficient and bearing equivalent thermal resistance through its excellent nonlinear estimation capabilities, incorporate them into the extended state-space model for prediction and updating, thereby tracking parameter changes in real time and correcting model deviations. By updating key parameters, the consistency between the digital twin model and the actual braking system is ensured, thereby suppressing pressure fluctuations and braking vibrations caused by parameter deviations.

[0031] The dynamic optimization problem is solved using the Pontryagin minimum principle or the real-time rolling optimization method to maximize energy recovery during the braking cycle while satisfying the system pressure stability constraints.

[0032] The attention mechanism calculates the weights associated with the temporal feature vector and the code feature vector, and the specific formula for generating the comprehensive feature vector is as follows: ; In the formula, Here, attention weights are used, and the score is a relevance scoring function. For time series feature vectors, For code feature vectors, This is a comprehensive feature vector.

[0033] Example 2: A hydraulic braking system for a horizontal machining center, based on Example 1, further includes a vibration suppression module and an adaptive damping adjustment unit. The vibration suppression module is equipped with a triaxial acceleration sensor and an adaptive notch filter. The vibration suppression module is used to monitor the vibration frequency and amplitude during the spindle braking process, and actively suppresses the resonance point by adjusting the compensation pulse width of the piezoelectric ceramic auxiliary valve.

[0034] The vibration suppression module collects vibration signals during braking through a triaxial acceleration sensor installed at the bottom of the spindle box. It then converts the time-domain signal into frequency-domain features using Fourier transform to identify the resonant frequency in the 10-500Hz frequency band. Subsequently, it controls the magnetorheological damper in the adaptive damping adjustment unit to change the magnetic field strength and adjust the damping coefficient in real time.

[0035] The adaptive notch filter uses the LMS algorithm to dynamically track the dominant vibration frequency. Its convergence factor μ is adaptively adjusted according to the vibration energy. When the amplitude is detected to exceed the threshold of 2.5μm, the filtering process with a notch depth of 40dB is automatically started. Combined with the vibration prediction model in the digital twin control module, the maximum amplitude during braking can be controlled within 1.2μm, reducing the surface texture defects of the workpiece caused by vibration.

[0036] The magnetorheological damper's built-in coil winding is made of tightly wound enameled copper wire, and its piston has 12 damping channels. When an adjustable current of 0-3A is applied to the coil, the magnetorheological fluid will complete the transformation from Newtonian fluid to Bingham fluid within 10ms under the action of the magnetic field, so that the damping force can be continuously adjusted within the range of 500-5000N, thereby suppressing the chatter caused by inertial impact during braking and reducing the vibration amplitude at the spindle end to below 0.02mm.

[0037] The implementation principle of this application embodiment is as follows: By fusing the LSTM neural network of the braking prediction module with the multimodal analysis of the CNC code parser, the braking start point, type, and pressure parameters are predicted. Combined with the reduced-order surrogate model and K-nearest neighbor algorithm of the digital twin control module, the pressure curve is initialized at the microsecond level, ensuring the timeliness of the braking response. At the same time, the coordinated work of the proportional servo valve and the piezoelectric ceramic auxiliary valve in the graded hydraulic valve group, along with the real-time optimization of the model predictive controller, can suppress pressure overshoot and fluctuations, improving the accuracy of braking pressure control. In addition, the micro variable displacement hydraulic motor / pump unit can recover and reuse hydraulic energy during braking, reducing system energy consumption. The combination of the thermo-mechanical coupling model and the online parameter identification module, through the extended Kalman filter, performs real-time recursive estimation and updating of key parameters, ensuring the dynamic consistency between the digital twin model and the actual system, and improving the system's adaptability under complex working conditions.

[0038] Furthermore, the application of the vibration suppression module and the adaptive damping adjustment unit, through the synergistic effect of the triaxial acceleration sensor, the adaptive notch filter and the magnetorheological damper, can identify and suppress the resonant frequency during the braking process, control the maximum amplitude at an extremely low level, reduce workpiece surface quality defects caused by vibration, and improve processing accuracy and product qualification rate.

[0039] The embodiments described in this specific implementation are preferred embodiments of this application and are not intended to limit the scope of protection of this application. Identical components are represented by the same reference numerals. Therefore, all equivalent changes made to the structure, shape, and principle of this application should be covered within the scope of protection of this application.

Claims

1. A hydraulic braking system for a horizontal machining center, comprising a braking prediction module, a graded hydraulic valve group, and a digital twin control module, characterized in that: The braking prediction module is used to analyze the time series of spindle speed, load current, and temperature data and predict the braking start point. The graded hydraulic valve group consists of a proportional servo valve and a piezoelectric ceramic auxiliary valve. The digital twin control module is based on a computational fluid dynamics simulation model and a PID controller to simulate the hydraulic circuit and generate the optimal pressure curve to control the graded hydraulic valve group. The braking prediction module integrates a spindle speed sensor, a load current detector, and a temperature probe. The spindle sensor, load current detector, and temperature probe are all fixed to the corresponding interfaces of the spindle housing with bolts. The proportional servo valve is connected to the hydraulic pump outlet pipeline through a flange. The piezoelectric ceramic auxiliary valve is fastened to the downstream branch of the main valve through threads. The industrial control computer of the digital twin control module is installed in the electrical cabinet through a bracket, and its signal lines are connected to each sensor and valve using shielded twisted-pair cables.

2. The hydraulic braking system for a horizontal machining center according to claim 1, characterized in that: The braking prediction module also includes a CNC code parser, which is used to parse the machining code segment to be executed in real time, extract geometric features and cutting parameters. The braking prediction module adopts a multimodal feature fusion network, which fuses the temporal feature vector output by the LSTM neural network with the code feature vector output by the CNC code parser, and performs multi-task learning.

3. The hydraulic braking system for a horizontal machining center according to claim 1, characterized in that: The digital twin control module uses intrinsic orthogonal decomposition or dynamic mode decomposition to reduce the dimensionality of the simulation model and obtain a surrogate model. Based on the surrogate model, the digital twin control module generates braking pressure curve plans for various working conditions offline and clusters the plan curves.

4. The hydraulic braking system for a horizontal machining center according to claim 3, characterized in that: During online control, the optimal pre-plan in the pre-plan library is quickly matched based on the current operating parameters using the K-nearest neighbor algorithm, achieving microsecond-level initialization of the pressure curve.

5. A hydraulic braking system for a horizontal machining center according to claim 3, characterized in that: The piezoelectric ceramic auxiliary valve is also equipped with a deformation micro-sensor. The feedback signal of the deformation micro-sensor is connected to the digital twin control module to sense the instantaneous viscosity change of the oil and realize cross-sensing compensation. A model predictive controller is also provided between the digital twin control module and the staged hydraulic valve group. The model predictive controller uses a surrogate model as the internal predictive model, solves the finite time domain optimization problem in each control cycle, obtains the optimal control sequence, and sends it to the PID controller for execution.

6. The hydraulic braking system for a horizontal machining center according to claim 1, characterized in that: The graded hydraulic valve group also includes a miniature variable displacement hydraulic motor / pump unit integrated into the hydraulic circuit. The miniature variable displacement hydraulic motor / pump unit is used to convert hydraulic energy into electrical energy for storage during braking and to assist in outputting power when the system needs to accelerate or compensate for pressure. The digital twin control module is also used to optimize the energy recovery and utilization of the miniature variable displacement hydraulic motor / pump unit according to the predicted processing cycle.

7. The hydraulic braking system for a horizontal machining center according to claim 1, characterized in that: The digital twin control module also includes a thermo-mechanical coupling model, which couples the thermal deformation model of the spindle and bearing with the mechanical model of the hydraulic braking system. The hydraulic braking system also includes an online parameter identification module, which recursively estimates and updates the key parameters in the digital twin model based on the actual collected vibration, temperature, and pressure data.

8. A hydraulic braking system for a horizontal machining center according to claim 3, characterized in that: The online parameter identification module uses an extended Kalman filter, which incorporates key model parameters such as hydraulic oil viscosity-temperature coefficient and bearing equivalent thermal resistance as part of the state variables to construct an extended state-space model. Through state prediction and update steps, the model parameters are recursively estimated and updated using actual observation data.

9. A hydraulic braking system for a horizontal machining center according to claim 1, characterized in that: It also includes a vibration suppression module and an adaptive damping adjustment unit. The vibration suppression module is equipped with a triaxial acceleration sensor and an adaptive notch filter. The vibration suppression module is used to monitor the vibration frequency and amplitude during the spindle braking process, and actively suppresses the resonance point by adjusting the compensation pulse width of the piezoelectric ceramic auxiliary valve.

10. A hydraulic braking system for a horizontal machining center according to claim 9, characterized in that: The vibration suppression module collects vibration signals during the braking process using a triaxial acceleration sensor installed at the bottom of the spindle box. It then converts the time-domain signal into frequency-domain features using Fourier transform to identify the resonant frequency in the 10-500Hz frequency band.

Images